I have a few doubts in the proof of existence of free groups given by Clara Loeh in page-22,23 of her Geometric Group theory book.

Theorem 2.2.7: Let S be a set. Then there exists a group freely generated by S. (By the previous proposition, this group is unique up to isomorphism.)

  1. When the equivalence is defined in page-22, is it implicitly assumed that $ s$ and $ \hat{s}$ are inverses?

$$\forall_{x,y \in A^*} \forall_{s \in S} xs \hat{s}y \sim y xy$$ $$\forall_{x,y \in A^*} \forall_{s \in S} x \hat{s} s y \sim xy$$

  1. In page-23, Clara defines a function which help us define inverse for general words. Here is what she gave exactly:

Inductively (over the length of sequences), we define a map $I:A^* \to A^* $ by $I(\epsilon) := \epsilon$ and

  1. $I(sx) :=I(x) \hat{s}$

  2. $I( \hat{s} x) : = I(x) s$

In the above why did she write 1. and 2.? Shouldn't just one of the equations be enough? My thinking is that in $A^*$ , $s$ and $\hat{s}$ are just elements in the set $A^*$.

Secondly, I am not sure if I understand this function in itself. Suppose I have a word $abc$ made up of letters $a,b,c$ then would it be correct to write $I(abc)= I(bc) \hat{a}$..? and then push it down till $I(abc)=\hat{c} \hat{b} \hat{a}$?

  1. In page-23 (again), I am confused at the part where she defines $\varphi^*$ which map from the set to words to a group. The idea can be understood by the diagram: enter image description here

For every group G and every map$ \varphi: S −→ G$ there is a unique group homomorphism $\varphi: F(S) −→ G$ such that $\overline{\varphi} ◦ i = \varphi$. Given $\varphi$, we construct a map

$\phi^*: A^* \to G$

Inductively by:

  1. $ \epsilon \to e$

  2. $sx \to \varphi(s) \cdot \varphi^*(x)$

  3. $ \hat{s} x \to ( \varphi(s))^{-1} \cdot \varphi^*(x)$

Again I am confused, why we need both two and three..? I think one is enough for similar reason said before.


  1. Free groups are defined by their universal property in the book.

1 Answer 1


For your first question: the proof of theorem $2.2.7$ explicitly explains what $\hat{s}$ is: it's an additional disjoint copy of $S$ whose elements play the roles of inverses. In fact, it's the very first paragraph of the proof.

For your second question: since $s\in S$ and $\hat s \in S$ are different (and belong to different, disjoint copies of $S$) we need to define how they both operate on the group element $x$, which is why both points $1$ and $2$ are given. You can indeed go from $I(abc)$ to $\hat{c}\hat b\hat a$ and this is also the expected right-inverse of the word $abc$ in a group.

And for your last question the answer is again that this explicitly connects the nominal inverse $\hat s$ with the inverse of the homomorphism $\varphi$ so both need to be defined.

  • $\begingroup$ If the group is not commutative then knowing what $x\hat s$ is tells you nothing about $\hat s x$. So you have to define both terms separately. Nominal inverse means "that thing we are calling an inverse" because: $S$ is set and sets don't have inverses. But if we can treat $S$ as a group, then it does, because groups are required to have inverses. Since groups are also closed, all the inverses $\hat s$ of elements $s\in S$ must also be in $S$, so we can take a second copy of $S$ and treat them as in the inverses. We're calling them inverses, but at this stage we haven't shown they are $\endgroup$
    – postmortes
    Apr 12 at 11:38
  • $\begingroup$ I don't really understand what you're writing there I'm afraid. $\hat s x$ and $x \hat s$ are different things, as Ms. Löh defines, and are on equal footing inasmuch as they are both words. $\endgroup$
    – postmortes
    Apr 12 at 11:55
  • $\begingroup$ Ah ok! I'm sorry, that's completely my fault; points one and two are not about commutativity (I won't bore you with the connection that made me write that); I've updated the answer to reflect that. The issue is that $s$ and $\hat s$ belong to different copies of $S$ so we need to specify how they both behave with respect to $I$. $\endgroup$
    – postmortes
    Apr 12 at 14:23
  • $\begingroup$ Could you explain right side of third eqtn? I am confused at $( \varphi(s))^{-1}$ $\endgroup$ Apr 12 at 14:41
  • 1
    $\begingroup$ If you have more questions you should really ask them as a new question and not keep adding them in comments. However: the point of $3.$ is to establish that $\phi ^*$ is a homomorphism and that it interacts with $\hat s$ by inverses (because we intend $\hat s$ to be the inverse of $s$). From $\epsilon = \phi(\varepsilon) = \phi(\hat s s) = \phi(\hat s) \phi(s) \Rightarrow (\phi(s))^{-1} = \phi(\hat s)$ (writing the homomorphism multiplicatively) being what we want we obtain $3$. $\endgroup$
    – postmortes
    Apr 13 at 6:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.